Custom AI - Python - Streamlit - MiniConda

This class you will learn how to:

• Set-Up a mini-conda environment
• deploy a Custom AI Instance

Set-Up MiniConda

You should do this as the user/domain folder which will be the host.

mkdir -p ~/miniconda3
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh -O ~/miniconda3/miniconda.sh
bash ~/miniconda3/miniconda.sh -b -u -p ~/miniconda3
rm -rf ~/miniconda3/miniconda.sh

After installing, initialize your newly-installed Miniconda. The following commands initialize for bash and zsh shells:

~/miniconda3/bin/conda init bash
~/miniconda3/bin/conda init zsh

Restart the shell

exec bash

Source the updated .bash configuration

source ~/.bashrc

Create your conda environment

conda create --name aiinstancename python=3.11

Activate your conda enviroment

conda activate aiinstancename

Install dependancies

pip install streamlit
pip install openai

Set up the System Service

This you need to do as root

su -

Create the service eg: aiinstance.service

[Unit]
Description=v1 client
After=network.target


[Service]
User=flastcom
WorkingDirectory=/home/xxxcom/xxxServers/aiinstance
ExecStart=/home/xxxcom/miniconda3/envs/aiinstance/bin/python -m streamlit run /home/xxxcom/xxxServers/aiinstance/app_anyscale_streamlit.workswithprompt.py --server.port 35003
Restart=always


[Install]
WantedBy=multi-user.target

Enable - Start - Status

systemctl enable aiinstance.service
systemctl start aiinstance.service
systemctl status aiinstance.service

You should see your aiinstance running like this

[root@server:system]$ systemctl status v1-lmw.client.service
? v1-lmw.client.service - v1 client
   Loaded: loaded (/etc/systemd/system/v1-lmw.client.service; enabled; vendor preset: disabled)
   Active: active (running) since Thu 2024-02-08 00:48:40 EST; 7s ago
 Main PID: 1822956 (python)
   CGroup: /system.slice/v1-lmw.client.service
           ??1822956 /home/xxxcom/miniconda3/envs/aiinstance/bin/python -m streamlit run /home/xxxcom/xxxServers/aiinstance/app_anyscale_streamlit.workswithprompt.py --server.port 35003


Feb 08 00:48:40 server.xxx.life systemd[1]: Started v1 client.
Feb 08 00:48:42 server.xxx.life python[1822956]: Collecting usage statistics. To deactivate, set browser.gatherUsageStats to False.
Feb 08 00:48:42 server.xxx.life python[1822956]:   You can now view your Streamlit app in your browser.
Feb 08 00:48:42 server.xxx.life python[1822956]:   Network URL: http://xxx.249.5.201:35003
Feb 08 00:48:42 server.xxx.life python[1822956]:   External URL: http://xxx.249.5.201:35003

Finally you need to use stunnel to push the http over https:

[File-AiInstance-domain]
accept = 35003  # this is the http port
connect = 35004 # this is the https port 
cert = /var/cpanel/ssl/apache_tls/xxx.com.au/combined  #these are the domains SSL certs locations 
key = /var/cpanel/ssl/apache_tls/xxx.com.au/combined